arXiv cs.CL· Jinyoung Park, Sanghyeok Lee, Omar Zia Khan, Hyunwoo J. Kim, Joo-Kyung Kim·· 11 小时前AI 评分24
HyperGraphPro：面向结构引导超图 RAG 的进度感知强化学习
HyperGraphPro: Progress-Aware Reinforcement Learning for Structure-Guided Hypergraph RAG
AI 导读
HyperGraphPro 是一个进度感知的超图 RAG 框架，将证据路径进展作为检索与策略优化的共同目标。检索端用查询条件的结构区分度对候选超边重排序，结合实体-查询相关性与局部-全局超图关联信号；策略优化端在轨迹级奖励之外加入步级进度感知优化，按答案引导的目标超边覆盖度分配步级奖励。多跳问答基准实验显示其推理准确率持续优于现有 GraphRAG 方法。
正文 · 原文
Abstract:Multi-hop question answering requires retrieval agents to construct a connected evidence path across reasoning turns rather than retrieve isolated relevant evidence for a single query. However, in existing GraphRAG methods, retrieval and reinforcement learning are often optimized with mismatched signals: retrieval is driven by local semantic relevance, while outcome-level reinforcement learning assigns the reward by assessing the final answer without identifying which retrieval actions progress the reasoning path. We introduce HyperGraphPro, a progress-aware hypergraph RAG framework that focuses on evidence-path progress as the shared objective for retrieval and policy optimization. For retrieval, HyperGraphPro reranks candidate hyperedges using query-conditioned structural distinctiveness, combining entity-query relevance with a local-vs-global hypergraph incidence signal to favor entities that are informative in the current graph neighborhood. For policy optimization, HyperGraphPro augments uniform trajectory-level reward assignment with stepwise progress-aware policy optimization, which assigns step-level rewards according to the coverage of answer-guided target hyperedges. Experiments on multi-hop question answering benchmarks demonstrate that HyperGraphPro consistently improves reasoning accuracy over existing GraphRAG methods.
| Comments: | In progress |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2601.17755 [cs.CL] |
| (or arXiv:2601.17755v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2601.17755 arXiv-issued DOI via DataCite |
Submission history
From: Jinyoung Park Dr. [view email]
[v1]
Sun, 25 Jan 2026 08:58:44 UTC (178 KB)
[v2]
Sun, 12 Apr 2026 14:35:33 UTC (138 KB)
[v3]
Thu, 5 Nov 2026 11:08:51 UTC (44 KB)
来源：arXiv cs.CL · arxiv.org